# Fileset

[20240730Cobbinah-MMTA.pdf](https://mdr.nims.go.jp/filesets/57d309d9-d59e-4539-914f-9f970ef9b873/download)

## Creator

Prince Valentine Cobbinah, Sae Matsunaga, [Yoshiaki Toda](https://orcid.org/0000-0002-8343-2890), Ryosuke Ozasa, Takuya Ishimoto, Takayoshi Nakano, Tsutomu Ito, Yoko Yamabe-Mitarai

## Rights

This version of the article has been accepted for publication, after peer review (when applicable) and is subject to Springer Nature’s AM terms of use, but is not the Version of Record and does not reflect post-acceptance improvements, or any corrections. The Version of Record is available online at: https://doi.org/10.1007/s11661-025-07759-8[In Copyright](http://rightsstatements.org/vocab/InC/1.0/)

## Other metadata

[On the Enhanced Creep Performance in Ti6246 Achieved Through Laser Powder Bed Fusion (LPBF) Processing](https://mdr.nims.go.jp/datasets/48ac0e87-9b80-46d0-92b4-dcdb8eb1002b)

## Fulltext

1 On the enhanced creep performance in Ti6246 achieved through Laser Powder Bed Fusion 1 (LPBF) processing 2  3 Prince Valentine Cobbinaha*, Sae Matsunagaa, Yoshiaki Todab, Ryosuke Ozasac, Takuya 4 Ishimotoc,d,  Takayoshi Nakanoc, Yoko Yamabe-Mitaraia** 5  6 aDepartment of Advanced Materials Science, Graduate School of Frontier Sciences, The 7 University of Tokyo, 5-1-5 Kashiwanoha, Kashiwa, Chiba 277-8561, Japan 8 bCenter for Basic Research on Materials, National Institute for Materials Science, 1-2-1 Sengen, 9 Tsukuba, Ibaraki, 305-0047, Japan 10 cDivision of Materials and Manufacturing Science, Graduate School of Engineering, Osaka 11 University, 2-1 Yamadaoka, Suita, Osaka 565-0871, Japan 12 dAluminium Research Center, University of Toyama, 3190 Gofuku, Toyama 930-8555, Japan 13  14 *Corresponding authors’ email: p.cobbinah22s@ams.k.u-tokyo.ac.jp 15 **Corresponding authors’ email: mitarai.yoko@edu.k.u-tokyo.ac.jp 16 Abstract 17 The high susceptibility of the Ti-6Al-2Sn-4Zr-6Mo wt.% (Ti6246) alloy to microstructural 18 changes stands as a challenge when processed by the laser powder bed fusion (LPBF) 19 technology. However, leveraging the capabilities of the LPBF process to successfully control the 20 microstructure (and/or crystallographic texture) of the Ti6246 could improve mechanical 21 properties, particularly at elevated temperatures. In this study, the creep performance (at 500 °C) 22 of Ti6246 fabricated from three different LPBF processing conditions and heat-treated (HT) at 23 885 °C were investigated. In the as-built state, all the LPBFed-Ti6246 exhibited columnar 24 microstructures with crystallographic lamellar-like microstructure (CLM), a near single crystal-25 like microstructure (SCM), and polycrystalline microstructure (PCM) textures, respectively. At 26 low applied stresses (100 – 300 MPa), diffusional creep was the dominant deformation 27 mechanism and its resistance depended on grain size. The reference β-forged-HT Ti6246, 28 characterized by large equiaxed grains, exhibited the lowest strain rate compared to the columnar 29 microstructure of SX1 (CLM)-HT, SX2 (SCM)-HT, and SX3 (PCM)-HT. Conversely, 30 dislocation slip governed deformation at high applied stresses (400 - 580 MPa) and its efficacy 31   2 depended on the α/β interfaces in the microstructures. Disjointed columnar grains in SX1 32 (CLM)-HT and the deformation of the polycrystalline grains in SX3 (PCM)-HT indicated that 33 the melt pool boundaries were unstable in the LPBFed-Ti6246. SX2 (SCM)-HT exhibited the 34 longest creep life due to the relatively stable melt pool boundaries and the near <001> SCM 35 crystallographic texture parallel to the applied stresses. Shallow dimples, cleavage facets, and the 36 observation of laser scan tracks characterized the fracture surfaces of the LPBFed-Ti6246. These 37 indicated that failure occurred by intergranular fracture resulting from the formation of creep 38 voids at the melt pool boundaries. 39  40 Keywords: LPBF; columnar microstructure; melt pool boundary stability; creep voids; diffusion 41 creep, dislocation creep  42  43 Introduction 44  45 The Ti-6Al-2Sn-4Zr-6Mo wt.% (Ti6246) alloy is an α + β alloy classified under the same 46 group as the Ti-6Al-4V wt.% (Ti64) alloy. One primary advantage of the Ti6246 alloy over the 47 Ti64 is its high susceptibility to microstructural changes, which allows a wide heat treatment 48 processing window. However, this high sensitivity of the microstructure of the Ti6246 could pose 49 a challenge when processed using the laser powder bed fusion (LPBF). This is because the LPBF 50 is an additive manufacturing (AM) process that involves local melting, rapid cooling, remelting 51 of already solidified layers, and a repeated heating cycle during fabrication 1. This repeated cycle 52 is well known to influence the microstructures of LPBF processed parts significantly 2.  53 So far, few studies on Ti6246 have involved the LPBF process 3-5.  Carrozza et al. 3 varied 54 input energy density and achieved fully dense Ti6246 using the LPBF process. The authors also 55 observed prior-β grains of columnar morphology and a martensitic microstructure, with higher 56 input energy density resulting in finer α′′needles in the as-built samples. Furthermore, the as-57 built samples exhibited substantial ductility but low yield strength. Conversely, post-heat 58 treatments markedly increased the strength of the LPBFed-Ti6246 at the expense of ductility 4. 59 Peng et al. 5 also observed the α′′martensitic microstructure in LPBFed-Ti6246. Heat treatment 60 at 650 °C significantly improved the hardness of the samples owing to the decomposition of the 61   3 α′′phase to the α and β phases. As reported in the studies mentioned above as well as many other 62 studies on titanium (Ti) α + β alloys such as Ti64 6, 7, mechanical behavior primarily depends on 63 formed microstructural features such as phase morphology, the type of precipitate that forms, 64 precipitate size, and distribution.  65 At elevated temperatures, mechanical behavior is also influenced by the synergistic effects of 66 temperature and applied stress on the Ti alloy. Creep resistance is an example of a high-67 temperature mechanical property strongly desired in structural materials (e.g. Ti6246) used for 68 aerospace applications. To the best of the authors’ knowledge, no previous study has paid 69 attention to the creep performance of Ti6246 processed by the LPBF. In our previous study 8, we 70 optimized the LPBF process parameters over a wide range and studied their effects on the 71 Ti6246. First, we observed the formation of 𝛼/𝛼′phases of varying sizes depending on the LPBF 72 process parameters. Also, the far ends of the investigated parameters i.e. samples with volumetric 73 energy density (VED) ≥  62.5 J/mm3 formed columnar microstructures whereas VEDs ≤  30 74 J/mm3 formed polycrystalline microstructures. As a progress of our research, the present study 75 evaluates the creep performance of the LPBFed-Ti6246 focusing on three different processing 76 conditions (further details are provided in Section 2). Aside from the different microstructures, 77 the three processing conditions were chosen because of the unique texture they exhibited. Laser 78 power of 360 W at 600 mm/s scan speed (VED = 75 J/mm3) produced alternating textures 79 termed crystallographic lamellar-like microstructure (CLM) texture, 300 W at 1000 mm/s (VED 80 = 50 J/mm3) formed a near-single crystal-like microstructure (SCM) texture, and 180 W at 1200 81 mm/s (VED = 25 J/mm3) resulted in a polycrystalline crystallographic microstructure-like 82 (PCM) texture. This study is the first to report such unique textures for Ti6246. For further 83 insights and comparison with the LPBFed-Ti6246, the present study also investigated the 84 conventional 𝛽-forged Ti6246. 85 In all, texture is an important property determinant factor that significantly influences creep 86 resistance. Hence serving as a strong motivation for the current study. 87  88 Materials and methods 89 Gas-atomized Ti-6Al-2Sn-4Zr-6Mo wt.% (Ti6246) powders with spherical morphologies 90 served as starting powders. Rectangular samples of height, 40 mm, and width 7 mm were 91   4 fabricated using the EOS M290 M LPBF equipment (EOS GmbH, Germany). Table 1 92 summarizes the LPBF process conditions used. 93  94 Table 1: Ti6246 LPBF process conditions 95 Sample name Process parameters Laser power, P (W) Scan speed, v (mm/s) Hatch distance, d (μm) Powder layer thickness, t (μm) VED (J/mm3) SX1 (CLM) 360 800 100 60 75 SX2 (SCM) 300 1000 50 SX3 (PCM) 180 1200 25  96 The energy inputs involved in the LPBF process were quantified as volumetric energy density 97 (VED) and defined as: 98  99                                                                       VED (J/mm3) = 𝑃𝑣𝑑𝑡     Furthermore, the bidirectional scan strategy along the X-axis with no rotations in subsequent 100 layers was used. Therefore, the scanning direction was in the XY plane, and the build direction 101 was parallel to the YZ plane.  102 A Ti6246 ingot β-forged at 1000 °C was also investigated and used as a reference alloy for 103 comparison with the LPBFed-Ti6246. 104 The LPBFed and β-forged Ti6246 were heat-treated (HT) at 885 °C for 1 hour and air-cooled. 105 The samples were then machined into the “dog-bone” shape, as illustrated in Figure 1. Table 2 106 summarizes the gauge diameter and gauge length of the creep samples. The R-type 107 thermocouples were attached to each sample to measure the testing temperature and the 108 respective elongation was measured using a linear gauge. To investigate the occurring transitions 109 of deformation mechanisms during the tensile creep test, the step creep test was performed at 500 110 °C starting with a 100 MPa stress in air condition.  The applied stress was then increased to 200, 111 300, 400, 500, and 580 MPa. The tensile creep test continued until fracture occurred under 580 112 MPa. Furthermore, the strain rate and applied stresses were analyzed using the classic Bird-113 Mukherjee-Dorn equation, and the prevalent deformation mechanisms were estimated. 114  115   5  116 Figure 1: Schematics of “dog-bone” shape for (a) SX1 (CLM)-HT, SX2 (SCM)-HT, and SX3 117 (PCM)-HT and (b) β-forged-HT Ti6246 used for creep tests. 118  119 Table 2. Gauge diameter and gauge length of creep samples. 120 Sample name Gauge diameter (mm) Gauge length (mm) SX1 (CLM)-HT 2.830 12.960 SX2 (SCM)-HT 2.832 12.991 SX3 (PCM)-HT 2.837 13.053 β-Forged-HT 6.017 30.001 The JEOL JSM-7200F scanning electron microscope (SEM) equipped with an electron 121 backscatter diffractometer (EBSD) and backscattered electron (BSE) detector was used to 122 characterize the microstructure of the heat-treated, cross-section, and fractured surfaces of all the 123 tested samples. Foremost, the building directions (YZ plane) of all the heat-treated samples and 124 their respective cross-sections after the creep testing were mounted, metallographically ground, 125 and polished, with the final stage of polishing involving 0.05 μm colloidal silica for 10 minutes. 126 The EBSD analysis covered minimum areas measuring 300 x 300 μm with a step size of 0.9 μm. 127 The EBSD data was analyzed using the TSL OIM 8 software. 128   6 Results 129 1.1 Microstructure and texture of the as-built Ti6246 130 Figure 2 (a) - (c) shows the SEM-BSE images of the as-built Ti6246. The as-scanned 131 microstructures of SX1 (CLM), SX2 (SCM), and SX3 (PCM) exhibited well-defined columnar  132  133 Figure 2: SEM-BSE images of the as-built LPBFed-Ti6246 (a) SX1 (CLM), (b) SX2 (SCM), (c) 134 SX3 (PCM) observed from the build direction (YZ plane), and (d) β-forged sample. 135 morphologies in the build direction. Although typical of the LPBF process, the continuous and 136 defined columnar growths observed in the present study differ from the few reported studies on 137 LPBFed-Ti6246 3-5. Apart from the obvious difference in process parameters, one possible 138 explanation could be the high laser power employed in the present study. Particularly in SX1 139 (CLM), the columnar growth occurred in an alternating sequence of thick and thin columnar 140 structures, as depicted in Figure 2 (a). The thick columnar structures occurred at the overlapping 141 ends of the melt pools whereas the thin secondary columnar structures formed and connected at 142 the centers. 143   7 The defined continuous thick growth at the overlapping ends suggests that during the layer 144 addition process, the localized melting sufficiently remelted the topmost parts of already 145 solidified layers or exceeded the powder layer thickness of 60 μm promoting the epitaxial growth 146 of existing grains. Concurrently, nucleated new grains at the overlapping ends of the new melt 147 pools also inherited the growth direction through competitive growth leading to the columnar 148 microstructure. 149 As already reported 8, the black phases in all the SEM micrographs were identified as the 150 𝛼 / 𝛼′ martensitic phase. Additionally, it is worth noting that the 𝛼′′ and 𝜔  phases were not 151 identified in any of the as-built samples. The sizes of the precipitates (𝛼/𝛼′phase) formed in SX1 152 (CLM) were observed to be extremely refined (in the nanoscale) while those formed in SX3 153 (PCM) were relatively coarse (microsized). In SX2 (SCM), the sizes were between those of 154 SX1(CLM) and SX3 (PCM). The observed precipitate size differences were mainly attributed to 155 the VEDs used for the respective builds. Readers are referred to our previous study for further 156 insights 8. On the other hand, the β-forged sample exhibited an equiaxed microstructure with an 157 average grain size of 138  49 μ m, as depicted in Figure 2 (d). Also, the precipitated 158 𝛼′martensitic phase were refined and of the nanoscale. 159 Figure 3 shows the β-phase reconstructed EBSD IPF maps of the as-built Ti6246. In fact, it 160 has been reported that crystallographic texture evolution in the LPBF process largely depends on 161 the melt pool shape particularly the depth or curvature of the melt pool bottoms 9. Analogous to 162 the casting process, the LPBF process forms a non-homogeneous nucleation at the  163   8  164 Figure 3: Reconstructed 𝛽-phase EBSD IPF maps of (a-1) and (a-2) crystallographic lamellar 165 microstructure-like (CLM) texture in SX1, (b-1) and (b-2) near single-crystal-like microstructure 166 (SCM) texture in SX2, and (c-1) and (c-2) polycrystalline microstructure-like (PCM) texture in 167 SX3. (YZ plane - build direction, XY plane - scanning direction). [(c-1) is reproduced under 168 terms of the Creative Commons CC BY 4.0 license 8. Copyright 2024, The Authors. Published 169 by Elsevier B.V. on behalf of KeAi Communications Co. Ltd.] 170 solid phase boundary at the melt pool bottom (or fusion region). The direction of heat flow is 171 fastest perpendicular to the solid-liquid interface, with grains competitively and at the same time 172 preferentially growing along in the opposite direction 10. The <111> and <001> textures were 173 observed to alternate in a lamellar-like pattern in the scanning direction (XY plane) of SX1 174 (CLM) fabricated with a 360 W laser power input at a scan speed of 800 mm/s. This lamellar-175 like orientation of crystallographic texture has been reported in other studies and is termed as 176 crystallographic lamellar-like microstructure (CLM) texture 11. The evolution of the CLM texture 177 is usually attributed to the formation of deep melt pools 11. The employed laser power input of 178 360 W at a slow speed (800 mm/s) combination (high energy) produced deep melt pools in SX1 179 (CLM) during the building process. The formed deep melt pools enhanced the epitaxial growth 180 of grains resulting in the columnar microstructure and promoted the inheritance of the existing 181 crystallographic orientations from one layer to another. According to Thijs et al. 12, builds 182 involving high-energy inputs are more likely to result in the formation of the <111> texture. On 183 the other hand, the <001> texture is seen to have evolved at the center of the melt pools where 184   9 energy is highest during the LPBF process. Thus, the <001> texture preferentially and 185 continuously grew at the center of the connecting melt pools along the build direction (YZ plane) 186 owing to the downward heat flow.  187 Similarly, a near <001> and <110> single crystal-like microstructure (SCM) texture formed in 188 the build (YZ plane) and scanning (XY plane) directions of SX2 (SCM) (300 W, 1000 mm/s) as 189 presented in Figures 3 (b-1) and (b-2), respectively. The columnar microstructure, as evident in 190 Figure 2 (b), suggests relatively deep melt pools also formed in SX2 (SCM). Gokcekaya et al. 13 191 postulated that when deep melt pools form with the solidification front almost horizontal at the 192 melt pool centerlines, two easy-growth directions perpendicular to each other from the melt pool 193 sides and center occur resulting in the <001> cubic texture evolution along the build direction. 194 From another perspective, the evolution of the <001> texture along the build direction may be 195 attributed to the fixed laser scanning direction employed. Without any rotations, the bidirectional 196 scanning along the X-axis resulted in a homogeneous thermal flux in each melt pool (and layer), 197 ultimately promoting grain preferential growth in the <001> direction. Similar texture has been 198 reported for face-centered cubic (fcc) based alloys such as nickel-based superalloys and stainless 199 steels 9, 11, 14.   200 Although the SEM-BSE image of SX3 (PCM) showed a columnar microstructure, the EBSD 201 analysis revealed the polycrystalline crystallographic microstructure (PCM) texture formed in 202 both the build (YZ plane) and scanning (XY plane) directions, as shown in Figure 3 (c-1) and (c-203 2) respectively. Unlike SX1 (CLM) and SX2 (SCM), the curvature of the bottom of the melt 204 pools formed in SX3 (PCM) (180 W, 1200 mm/s) significantly decreased owing to both the low 205 laser power input and the fast scan speed. This combination of process parameters prevented the 206 epitaxial growth of grain at the fusion region (or melt pool bottom) and the consequent inheriting 207 of previous crystal orientations. Thus culminating in the observed randomly oriented small 208 grains in SX3 (PCM). 209  210 1.2 Microstructure of the heat-treated LPBFed-Ti6246 211 Figure 4 shows the SEM-BSE images of the heat-treated (HT) LPBFed and β-forged Ti6246. 212 As expected, the 𝛼′ martensitic phases coarsened and decomposed to the 𝛼  phase. This is 213 because, generally, the decomposition of the 𝛼′martensitic phase to the 𝛼 phase is diffusion-214 driven, requiring elevated temperature and sufficient time. Foremost, the micrographs provided 215   10 insight into the locations of the precipitated martensitic phase before heat treatment. The 𝛼 216 phase, therefore, were found within the columnar structures and boundaries (longitudinal 217 boundaries) and the melt pool boundaries. Similarly, in the β-forged-HT Ti6246, the 𝛼 phase 218 were located in the equiaxed grains as well as at the grain boundaries. This is in good agreement 219 with available literature on the Ti6246 15, 16.   220  221 Figure 4: SEM-BSE images of the heat-treated (HT) LPBFed-Ti6246 (a) SX1 (CLM), (b) SX2 222 (SCM), (c) SX3 (PCM) observed from the build direction (YZ plane), and (d) β-forged sample. 223 Figure 5 presents the volume fraction and average lengths of the 𝛼 phase in the LPBFed and 224 the β-forged Ti6246 after HT. The volume fraction of the 𝛼 phase in SX2 (SCM)-HT and SX3 225 (PCM)-HT was about the same (~ 40) after heat treatment. Conversely, the measured average 226 lengths increased from 3.78 μm in SX1 (CLM)-HT to 7.03 μm in SX3 (PCM)-HT. The β-forged-227 HT Ti6246, on the other hand, exhibited the least 𝛼 volume fraction (27), but the measured 228 average length (5.5 μm) was between those of SX2 (SCM)-HT and SX3 (PCM)-HT. 229  230   11  231 Figure 5: 𝛼 phase volume fraction and average length (𝜇m) in the LPBFed and the 𝛽-forged 232 Ti6246 after heat treatment (HT) at 885 °C for 1 hour and air cooled. 233 1.3 Creep behavior of the heat-treated LPBFed and 𝛽-forged Ti6246  234 Figure 6(a) shows the LPBFed and β-forged Ti6246 creep curves. Compared to the HT 235 LPBFed-Ti6246, the β-forged-HT Ti6246, with an equiaxed microstructure, exhibited the least 236 strain till fracture. Despite the similarity in the microstructures of LPBFed-Ti6246, their creep 237 lives varied, in the range of 380 – 440 hours. Overall, SX2 (SCM)-HT had the longest creep life. 238 Therefore, to determine the deformation mechanisms in the respective samples, step creep tests 239 were performed. Figure 6(b) shows the respective creep curves of the step creep tests. The 240 steady-state strain rate  241  242  243  244   12  245 Figure 6: (a) Overall and (b) step creep curves of SX1 (CLM)-HT, SX2 (SCM)-HT, SX3 (PCM)-246 HT, and 𝛽-Forged-HT Ti6246 at 500 °C. 247 (𝜀̇) usually varies with the applied stress (𝜎), the absolute temperature (𝑇), and the grain size of 248 the material (𝑑𝑔) through a relationship that can be expressed in the form of Eq. 1 17.   249  250 𝜀̇ = 𝜀𝑜̇ 𝐺𝑉𝑎𝑘𝑜𝑇 (𝑏𝑑𝑔)𝑝(𝜎𝐺)𝑛 𝐷𝑏2                                                 (1)  251 where 𝜀̇, 𝐺, 𝑉𝑎, 𝑘, 𝑏, 𝐷, 𝑝, and 𝑛 represent the strain rate independent of temperature and stress, 252 shear modulus, atomic volume, Boltzmann’s constant, Burgers vector, diffusion coefficient (= 𝐷0 253 exp (-𝑄 𝑅𝑇⁄ ), where 𝐷0  is a frequency factor, 𝑄  is the activation energy and 𝑅  is the gas 254 constant), the inverse grain size exponent, and apparent stress exponent, respectively. In the 255 present study, the grain size (i.e. melt pool size) and testing temperature do not change, thus the 256 steady-state strain rate (𝜀̇) can be considered as a function of the applied stress (𝜎). Therefore, 257 Eq. 1 can be further expressed as:  258  259 log 𝜀̇ = log 𝐴 + 𝑛 log 𝜎                                                       (2)  260 where 𝐴 = 𝜀𝑜̇ 𝐺𝑉𝑎𝑘𝑜𝑇 (𝑏𝑑𝑔)𝑝𝐷𝑏2     261 In general, the deformation mechanisms in metallic alloys that control high-temperature creep 262 include diffusional creep 18, 19, grain boundary or interface sliding 20, 21, dislocation glide and 263 climb 22, 23, and viscous dislocation glide controlled by dragging of solute atoms or by jogs 24, 25. 264   13 One method of knowing which deformation mechanism dominates is by determining the values 265 of the apparent stress exponent (𝑛 ). The gradient from plotting log 𝜀̇  vs log 𝜎 , is used to 266 determine the apparent stress exponent (𝑛) as depicted in Figure 7. For each applied stress, the 267  268  269 Figure 7: Dependence of steady-state creep on applied stress at 500 °C in (a) SX1 (CLM)-HT, (b) 270 SX2 (SCM)-HT, (c) SX3 (PCM)-HT, and (d) 𝛽-Forged-HT Ti6246. 271 minimum creep rate value was determined from the steady-state regime where the creep curve 272 exhibited a distinct plateau. As Figure 7 shows, the values of 𝑛 fall along two different lines 273 depending on the applied stresses in all the HT LPBFed and the β-forged Ti6246. The apparent 274 stress exponent ( 𝑛 ) increased rapidly when the applied stresses exceeded 300 MPa. This 275 breakaway indicates a transition in the occurring deformation mechanisms. It is well-established 276 that when 𝑛  1, the dominant creep process is associated with the Nabarro-Herring-Coble creep 277 (i.e. diffusion creep) mechanisms whereas when 𝑛  2-3 the grain boundary sliding mechanism 278 dominates. Above 3, dislocation slip is the dominant deformation mechanism. 279   14 The cross-section of each sample was investigated after fracture to understand the effect of the 280 various microstructures on the observed creep behaviors. The microstructures of the cross-281 section of the samples after the creep tests are shown in Figure 8. It was observed that the grains 282 in SX1 (CLM)-HT changed from long pronounced columnar structures to disjointed columnar 283 structures  284  285  286 Figure 8: SEM-BSE images of the cross-sections of (a) SX1 (CLM)-HT, (b) SX2 (SCM)-HT, (c) 287 SX3 (PCM)-HT, and (d) 𝛽-Forged-HT samples after fracture. The yellow dotted lines indicate 288 the grain boundaries. The elongated substructures were also marked with dotted lines. 289 (Figure 8(a)). On the other hand, samples SX2 (SCM)-HT and SX3 (PCM)-HT maintained their 290 columnar structures parallel to the build direction as depicted in Figure 8 (b) and (c), 291 respectively. Interestingly, the width of the columnar structures widened in SX3 (PCM)-HT from 292 92 μm to 268 μm. The threefold increase in the width may suggest the fusing of columnar 293 structures and is corroborated by the formation of elongated substructures (deformed grains) 294 noticed within the thick columnar structures. For the β-forged Ti6246, the equiaxed grains were 295 elongated in the direction of the applied stresses as depicted in Figure 8 (d).  296   15 1.4 Fracture surfaces after creep tests 297 SEM secondary electron images of the fractured surfaces after the creep tests are shown in 298 Figure 9. Many shallow dimples were observed in SX1 (CLM)-HT as depicted in Figure 9 (a). A 299 closer examination of the interior of the dimples revealed the existence of rings at different 300 depths (inset of Figure 9 (a)). These rings are likely melt pool boundaries from different layers of 301 the build. The fracture surface of SX2 (SCM)-HT also exhibited similar shallow dimples as 302 noticed in 303  304  305 Figure 9: SEM secondary electron images of the fracture surfaces of (a) SX1 (CLM)-HT, (b) 306 SX2 (SCM)-HT, (c) SX3 (PCM)-HT, and (d) 𝛽-Forged-HT Ti6246. 307 SX1 (CLM)-HT. In addition, clear outlines of the laser scan pathways were observed as indicated 308 by the dotted lines in Figure 9 (b). This observation suggests delamination i.e. separation of 309 layers likely from the applied constant pull. Furthermore, the polycrystalline SX3 (PCM)-HT 310 showed outlines of the laser scan pathways and some cleavage facets as depicted in Figure 9 (c). 311 The β-forged-HT Ti6246 exhibited a mix of secondary cracks, tearing edges, and cleavage facets 312 (Figure 9 (d)). At high magnification, micro-voids were observed at the grain boundaries as 313 arrowed in the inset of Figure 9 (d).  314   16 Although some obvious cleavage facets and tearing edges were observed in some samples, the 315 overall creep fracture surface morphologies of the LPBFed and β-forged-HT Ti6246 primarily 316 indicate intergranular brittle fracture modes. Similar observations have been reported by He et al. 317 26 on the creep performance of LPBFed-Ti64.  318 Discussion 319 Generally, the relatively low mechanical properties of additive manufactured (AM) products 320 are attributed mostly to defects such as pores. These pores are the vulnerable areas where cracks 321 initiate and propagate leading to failure. However, the LPBF AM process boasts of forming fully 322 dense products that help circumvent the effects of pores on mechanical properties, particularly at 323 room temperature. Aside from pores, many factors can influence the mechanical properties of 324 LPBFed products. The present study delved into uncovering and understanding the deformation 325 mechanisms of LPBFed-Ti6246 during creep at 500 °C.  326 1.5 Effect of microstructural stability on creep 327 The apparent stress exponent (𝑛) values for the applied stresses from 100 - 300 MPa in all the 328 Ti6246 samples indicate the diffusional creep mechanism controlled the creep process, as shown 329 in Figure 7. The mechanisms of diffusional creep test the stability of microstructures owing to 330 the movement of matter. As such, it is well established that diffusional creep rate hinges on 331 microstructural features including grain size and morphology 27. As evident in Figure 6, the β-332 forged-HT Ti6246 with equiaxed grains of average size 138  49 μm exhibited a lower strain rate 333 than the columnar structures of SX1 (CLM)-HT, SX2 (SCM)-HT, and SX3 (PCM)-HT. The 334 observed low strain rate implies that the large grain sizes of the β-forged-HT Ti6246 showed 335 resistance to diffusional creep owing to the increased distance through which matter had to 336 diffuse for deformation to occur. In addition, we argue that the equiaxed morphology and large 337 grain sizes in the β-forged-HT Ti6246 presented a more stabilized microstructure because the 338 microstructure homogenously deformed under low applied stresses. In other words, the 339 elongation of the individual grains was directly proportional to the macroscopic elongation of the 340 β-forged-HT Ti6246 sample (Figure 8(d)). 341 In the case of the HT LPBFed-Ti6246, because the columnar structures were parallel to the 342 applied stresses, the effects of the diffusional creep manifested at the melt pool and longitudinal 343   17 grain boundaries. Although no significant differences were observed in the strain rates in SX1 344 (CLM)-HT, SX2 (SCM)-HT, and SX3 (PCM)-HT, the melt pool boundaries perpendicular to the 345 applied stresses were the most vulnerable regions as a result of the nonuniform distribution of 346 stress concentration as creep proceeded 28. Hence microstructural stability in the HT LPBFed-347 Ti6246 during diffusional creep depended on the stability of the melt pool boundaries. This is 348 because the melt pool boundaries act as conduits for atom migration during diffusional creep. 349 One effect of the melt pool boundary instability as diffusional creep occurred, is seen in SX1 350 (CLM)-HT (Figure 8(a)) wherein the long pronounced columnar structures changed to disjointed 351 columnar structures. In SX3 (PCM)-HT, the effect of the melt pool boundary instability, as 352 already mentioned, culminated in elongated substructures (deformed grains) and the increase in 353 the size of the columnar structures. Consistent with literature 29, diffusion enabled the melt pool 354 boundaries in the polycrystalline microstructure of sample SX3 (PCM)-HT to slide as an 355 accommodation mechanism. The sliding of the melt pool and longitudinal boundaries preserved 356 the continuous deformability while avoiding microcracking during diffusion creep. On the other 357 hand, compared to the SX1 (CLM)-HT, SX3 (PCM)-HT, and the β-forged-HT Ti6246, the 358 microstructure of SX2 (SCM)-HT showed no noticeable changes from the mechanisms of 359 diffusional creep. We can, therefore, infer that the melt pool and longitudinal boundaries were 360 stable and as such efficiently resisted the diffusional deformation at the low applied stresses.  361 1.6 Role of the 𝛼/β interface  362 It is well documented that precipitates such as the 𝛼  phase inhibit diffusional creep 30. 363 However, as the applied stresses increased above 300 MPa, the deformation mechanism in all the 364 HT LPBFed and β-forged Ti6246 deviated from diffusional creep to dislocation slip (with 𝑛  365 2.3 (SX1 (CLM)-HT), 3.8 (SX2 (SCM)-HT), 3.1 (SX3 (PCM)-HT), and 3.5 (β-forged-HT)). As 366 elucidated in our previous study 8, the precipitate sizes in the as-built Ti6246 were quite peculiar. 367 Contrary to several existing studies 31, 32, the low laser power input at fast scan speed 368 combination (fast solidification rate) used in SX3 (PCM) resulted in coarse martensitic 369 precipitates whereas the high laser power input at slow scan speed (slow solidification rate) in 370 SX1 (CLM) formed refined martensitic precipitates.  Figure 10 compares the 𝛼-phase volume 371 fraction and average lengths after heat treatment (HT) and creep tests at 500 °C. The volume 372 fraction of the 𝛼 phase after HT and creep tests remained nearly the same in SX1 (CLM)-HT, 373   18 SX2 (SCM)-HT, and SX3 (PCM)-HT as well as the β-forged-HT. Furthermore, the 𝛼-phase 374 volume fraction in the  375  376  377 Figure 10: 𝛼 phase volume fraction and average length (𝜇m) in the HT LPBFed and the 𝛽-378 forged-Ti6246 before and after the creep test at 500 °C. 379 respective samples constituted less than 50% of the overall microstructure, indicating that the 380 minimum creep rate at stresses greater than 300 MPa depended on the deformation of the 𝛼-381 phase precipitates and the β-phase matrix. Since the 𝛼-phase precipitates and the β-phase matrix 382 exhibit different strengths, their interface experiences all the occurring local strains. During 383 deformation, these 𝛼 / β  interfacial regions become barriers to dislocations, and as a result, 384 account for the deformation resistance under high applied stresses. However, the more 385 dislocations accumulate and interact energy increases, which simultaneously decreases the 386 efficacy of the 𝛼/β interfacial regions. The role of the 𝛼/β interfacial regions suggest the 𝛼-phase 387 volume fraction has a direct relation to deformation resistance at high applied stresses. This was 388 made evident in the β-forged-HT Ti6246, where the low 𝛼-phase volume fraction culminated in a 389 large accelerated deformation at stresses greater than 300 MPa (Figure 6(a)). Along similar lines, 390 Ro et al. 33 after investigating several 𝛼/𝛽 Ti alloys at 500 °C, asserted that the presence of a high 391 𝛼-phase volume fraction tends to decrease the steady-state creep rate in good agreement with the 392 observed deformation behavior in SX1 (CLM)-HT, SX2 (SCM)-HT, and SX3 (PCM)-HT. In all, 393   19 despite being a classic concept, the notion that the 𝛼/β interfacial regions control deformation 394 resistance at high applied stresses remains valid, as shown in this study.  395 1.7 Reasons for the enhanced creep performance in SX2 (SCM)-HT 396 Despite the similarities in creep behavior in the HT LPBFed-Ti6246, sample SX2 (SCM)-HT 397 had the longest creep life. One reason might be the relatively stable microstructure (melt pool 398 boundaries) observed in Figure 8(b). Indeed, the columnar microstructure of SX2 (SCM)-HT did 399 not change or form any substructures except for the elongation of the 𝛼 needles. From Figure 10, 400 the 𝛼 needles in SX2 (SCM)-HT showed the longest elongation. This suggests SX2 (SCM)-HT 401 experienced the highest plastic deformation and the highest resistance to dislocation movements 402 34. However, another possible explanation might lie beyond the sample's microstructure i.e. 403 texture. Texture is a material property determinant known to influence creep significantly. As 404 previously shown, the as-built texture in SX2 (SCM)-HT suggests that the crystallographic 405 direction of the solidifying β-phase aligned in the <001> direction, parallel to the build direction 406 and the applied external tensile stresses as presented in Figure 3 (b-1). This SCM texture perhaps 407 accounts for the improved creep life performance. The <001> texture is the easy growth 408 direction (preferential growth direction) of the body-centered cubic (bcc) β-phase and usually 409 requires less energy for growth 35. Moreover, the <001> direction possesses the least atom 410 arrangements, resulting in the lowest elastic modulus 36. Therefore, when high tensile stresses 411 were applied to SX2 (SCM)-HT, the β-phase matrix was subjected to an increased strain owing 412 to the smaller stiffness than that of the 𝛼 needles 36. The effect from the strained <001> matrix 413 together with the ability of the 𝛼  needles to hinder dislocation glide explains the observed 414 improved creep life.  415 1.8 Creep voids formation, growth, and failure 416 In SX1 (CLM)-HT, SX2 (SCM)-HT, and SX3 (PCM)-HT, the melt pool and longitudinal 417 boundaries constituted weak bonds. These weak bonds caused the nonuniform stress distribution 418 at the boundaries during diffusional creep, leading to sliding and forming creep voids. As the 419 applied stress increased, growth of the creep voids was driven by dislocation motion around the 420 voids which ultimately resulted in the pull-out manifested as rings or shallow dimples, and the 421 observed outlines of the laser scan pathways. The schematic presented in Figure 11 illustrates the 422 creep voids initiation, and growth processes leading to fracture in the HT LPBFed Ti6246.  423   20  424  425 Figure 11: Schematic of creep voids formation, growth, and fracture of the HT LPBFed Ti6246. 426  427 Additionally, it is worth noting that the number of creep voids increased with the 𝛼-phase 428 volume fraction in the respective samples, as evident in Figures 9 and 10. In all, the failure of the 429 HT LPBFed-Ti6246 (SX1 (CLM)-HT, SX2 (SCM)-HT, and SX3 (PCM)-HT) was caused by the 430 formation of creep voids that grew into microcracks at the melt pool boundaries (or grain 431 boundaries) leading to intergranular fracture. 432 Conclusion 433 In this study, the microstructure, crystallographic texture, and high-temperature creep 434 performance of Ti-6Al-2Sn-4Zr-6Mo wt.% (Ti6246) fabricated using three different LPBF 435 process conditions (SX1 - 360 W, 800 mm/s; SX2 – 300 W, 1000 mm/s; SX3 – 180 W; 1200 436 mm/s) were investigated. Creep tests on the heat-treated (HT) LPBFed-Ti6246 were carried out 437 at 500 °C. To understand occurring deformation mechanisms, the step creep test was first 438 performed for each sample. The following conclusions were drawn:  439  The as-built LPBFed-Ti6246 exhibited columnar microstructures but with three unique 440 crystallographic textures. SX1 formed a crystallographic lamellar-like microstructure 441 (CLM) texture while SX2 and SX3 formed the near <001> single crystal-like 442 microstructure (SCM) and polycrystalline microstructure (PCM) texture, respectively. 443   21  444  Heat treatment (HT) at 885 ° C provided sufficient energy and time for the 445 𝛼/𝛼′martensitic phase to decompose to the 𝛼 phase and grow. The 𝛼 phases were located 446 within the columnar structures and the melt pool boundaries of SX1 (CLM)-HT, SX2 447 (SCM)-HT, and SX3 (PCM)-HT. 448  At low applied stresses (100 – 300 MPa), diffusional creep was the dominant deformation 449 mechanism and its resistance depended on the grain size of the tested samples. The β-450 forged-HT, characterized by large equiaxed grains (138  49 μm), exhibited the lowest 451 strain rate compared to the columnar microstructure of SX1 (CLM)-HT, SX2 (SCM)-HT, 452 and SX3 (PCM)-HT. 453  Conversely, dislocation slip governed deformation at high applied stresses (400 - 580 454 MPa) and its efficacy depended on the 𝛼/β interfaces in the microstructures to impede 455 dislocation mobility.  456  The observed disjointed columnar grains in SX1 (CLM)-HT and the deformation of the 457 polycrystalline grains in SX3 (PCM)-HT indicated that the melt pool boundaries were 458 unstable in the LPBFed-Ti6246. 459  Nonetheless, SX2 (SCM)-HT exhibited the longest creep life. Compared to SX1 (CLM)-460 HT and SX3 (PCM)-HT, SX2 (SCM)-HT maintained its continuous columnar structure 461 till fracture. The relatively stable melt pool boundaries coupled with the near <001> SCM 462 crystallographic texture of the β matrix in the build direction and parallel to the applied 463 stresses culminated in an improved creep life compared to the reference β-forged-HT 464 Ti6246.  465  Shallow dimples, the observation of laser scan tracks, and cleavage facets characterized 466 the fracture surfaces of the SX1 (CLM)-HT, SX2 (SCM)-HT, and SX3 (PCM)-HT. The 467 failure mode in the SX1 (CLM)-HT, SX2 (SCM)-HT, and SX3 (PCM)-HT was by 468 intergranular brittle fracture from forming creep voids, particularly at melt pool 469 boundaries perpendicular to the applied tensile stresses. 470  471 In all, the present study showed that the LPBF process can usefully influence the 472 microstructure and crystallographic texture of the Ti6246 alloy to attain enhanced properties. As 473 evident in SX2 (SCM)-HT, forming a single crystal-like microstructure texture can improve 474   22 creep performance. In other words, this study reinforces the concept that crystallographic texture 475 could be deliberately controlled using the LPBF process while simultaneously building parts 476 with sophisticated geometries. This serves to benefit the aerospace industry in terms of saving 477 time, energy, and cost in producing needed parts.  478 Declaration of Competing Interest 479 On behalf of all authors, the corresponding author states that there is no conflict of interest.  480 Acknowledgement 481 Special thanks to Mrs. Nina Kobata of NIMS for all her assistance during the creep tests. 482 Data availability 483 Data will be made available on request. 484 Funding 485 This work was supported by Grants-in-Aid for Transformative Research Area A [grant number 486 JP21H05198] and for Scientific Research [grant number JP23H00235] from the Japan Society 487 for the Promotion of Science (JSPS), and by The Light Metal Educational Foundation (Japan). 488 References 489 1. P. V. Cobbinah, R. A. Nzeukou, O. T. Onawale and W. R. Matizamhuka, Metals, 2020, 490 11, 58. 491 2. T. DebRoy, H. L. Wei, J. S. Zuback, T. Mukherjee, J. W. Elmer, J. O. Milewski, A. M. 492 Beese, A. Wilson-Heid, A. De and W. Zhang, Progress in materials science, 2018, 92, 493 112-224. 494 3. A. Carrozza, A. Aversa, P. Fino and M. Lombardi, Journal of alloys and compounds, 495 2021, 870, 159329. 496 4. A. Carrozza, A. Aversa, P. Fino and M. Lombardi, Materials & design, 2022, 215, 497 110512. 498 5. H. Peng, S. Wu, W. H. Kan, S. C. V. Lim, Y. Zhu and A. Huang, Scripta Materialia, 2023, 499 226, 115209. 500 6. P. Yadav and K. K. Saxena, Materials Today: Proceedings, 2020, 26, 2546-2557. 501   23 7. C. Veiga, J. P. Davim and A. J. R. Loureiro, Rev. Adv. Mater. Sci, 2012, 32, 133-148. 502 8. P. V. Cobbinah, S. Matsunaga, Y. Toda, R. Ozasa, M. Okugawa, T. Ishimoto, Y. Liu, Y. 503 Koizumi, P. Wang, T. Nakano and Y. Yamabe-Mitarai, Smart Materials in Manufacturing, 504 2024, 2, 100050. 505 9. K. Hagihara and T. Nakano, JOM, 2021, 1-14. 506 10. W. Shifeng, L. Shuai, W. Qingsong, C. Yan, Z. Sheng and S. Yusheng, Journal of 507 Materials Processing Technology, 2014, 214, 2660-2667. 508 11. O. Gokcekaya, T. Ishimoto, S. Hibino, J. Yasutomi, T. Narushima and T. Nakano, Acta 509 materialia, 2021, 212, 116876. 510 12. L. Thijs, M. L. Montero Sistiaga, R. Wauthle, Q. Xie, J.-P. Kruth and J. Van Humbeeck, 511 Acta materialia, 2013, 61, 4657-4668. 512 13. O. Gokcekaya, N. Hayashi, T. Ishimoto, K. Ueda, T. Narushima and T. Nakano, Additive 513 manufacturing, 2020, 36, 101624. 514 14. T. Ishimoto, S. Wu, Y. Ito, S.-H. Sun, H. Amano and T. Nakano, ISIJ international, 2020, 515 60, 1758-1764. 516 15. S. L. Corre, R. Forestier, F. Brisset, M. H. Mathon and D. Solas, in Proceedings of the 517 13th World Conference on Titanium, Wiley Online Library, 2016, pp. 757-764. 518 16. I. Weiss and S. L. Semiatin, Materials science & engineering. A, Structural materials : 519 properties, microstructure and processing, 1998, 243, 46-65. 520 17. T. G. Langdon, Metallurgical and Materials Transactions A, 2002, 33, 249-259. 521 18. K. Sotoudeh and P. S. Bate, Acta Materialia, 2010, 58, 1909-1920. 522 19. C. Herring, Journal of Applied Physics, 1950, 21, 437-445. 523 20. Y. Yamabe-Mitarai, T. Inoue, T. Kuroda, S. Matsunaga, Y. Toda, T. Matsunaga, T. Ito, R. 524 Ozasa, T. Ishimoto and T. Nakano, Materials Transactions, 2023, 64, 1175-1182. 525 21. E. Alabort, P. Kontis, D. Barba, K. Dragnevski and R. C. Reed, Acta Materialia, 2016, 526 105, 449-463. 527 22. K. R. Athul, U. T. S. Pillai, A. Srinivasan and B. C. Pai, Advanced Engineering Materials, 528 2016, 18, 770-794. 529 23. F. Appel, U. Lorenz, M. Oehring, U. Sparka and R. Wagner, Materials Science and 530 Engineering: A, 1997, 233, 1-14. 531   24 24. T. Suzuki, S. Takeuchi and H. Yoshinaga, Dislocation dynamics and plasticity, Springer 532 Science & Business Media, 2013. 533 25. F. A. Mohamed and T. G. Langdon, Acta metallurgica, 1974, 22, 779-788. 534 26. Y. He, Y. e. Ma, W. Zhang and Z. Wang, Engineering Failure Analysis, 2022, 135, 535 106063. 536 27. G. Malakondaiah and R. Rao, Defence Science Journal, 1985, 35, 201-217. 537 28. Z. H. Xiong, S. L. Liu, S. F. Li, Y. Shi, Y. F. Yang and R. D. K. Misra, Materials Science 538 and Engineering: A, 2019, 740-741, 148-156. 539 29. P. C. Millett, T. Desai, V. Yamakov and D. Wolf, Acta Materialia, 2008, 56, 3688-3698. 540 30. J. E. Harris, Metal Science Journal, 1973, 7, 1-6. 541 31. W. Huang and Y. Zhang, Journal of Manufacturing Processes, 2019, 42, 139-148. 542 32. J. Sun, X. Zhu, L. Qiu, F. Wang, Y. Yang and L. Guo, Materials Today Communications, 543 2019, 19, 277-285. 544 33. Y. Ro, S. Nakazawa, H. Onodera, K. Ohno, T. Yamagata, I. Tomizuka and M. Yamazaki, 545 ISIJ International, 1989, 29, 165-170. 546 34. R. Ullah, J. Lu, L. Sang, Y. Xiaoxiao, W. Zhang, Y. Zhang and Z. Zhang, Journal of 547 Alloys and Compounds, 2020, 817, 152781. 548 35. T. Ishimoto, K. Hagihara, K. Hisamoto, S.-H. Sun and T. Nakano, Scripta Materialia, 549 2017, 132, 34-38. 550 36. S. G. Tian, H. H. Zhou, J. H. Zhang, H. C. Yang, Y. B. Xu and Z. Q. Hu, Materials 551 Science and Technology, 2000, 16, 451-456. 552  553